TRAJECTORY PLANNING FOR

ROBOT MANIPULATORS

Prof. Rohan Munasinghe

Based on MSc Research by Chinthaka Porawagama

Industrial Robotics Involves in

Pick-and-place operations

Assembling operations

Loading and stacking

Automated welding, etc.

Proper motion planning is

needed in these applications
Trajectory Planning

Manipulators with multi degree of

freedom for accomplishing various
complex manipulation in the work
space

Path: only geometric description

Trajectory: timing included

Trajectory Planning
Joint Space Vs Operational Space

Planning in Operational Space

Planning in Operational Space

Planning in Joint Space

Path Definition
“Expressing the desired positions of a manipulator
in the space, as a parametric function of time”

Task to Trajectory
 Types of Motion
1. Point to point motion:
start
 End effector moves from a start point to
end point in work space
 All joints’ movements are coordinated for
the point-to-point motion end
 End effector travels in an arbitrary path

2. Motion with Via Points start

 End effector moves through an

intermediate point between start and end
 End effector moves through a via point
end via
without stopping

Joint Space Planning

Point to point motion:
“Describing of joints’ motions from start to end by smooth functions

End point ‘B’

θ1B θ2B θ3B θ3B
Basic stages of solving
θ2 θ3(t) of joint space trajectory
θ3 θ3A
planning problem ?
θ1
θ2B θ2(t)

θ2A
Start point ‘A’
θ1A θ2A θ3A θ1(t)
θ1B
θ1A
Arbitrary path in
the task space
A
Travel time T

Task Space Parametric Representation

Joint Space Planning
Point to point motion:
“Describing of joints’ motions from start to end by smooth functions

End point ‘B’

θ1B θ2B θ3B θ3B
1. Inverse kinematics of start
θ2 θ3(t)
θ3 θ3A and end points (A & B)

θ1
θ2B θ2(t)

2. Joint angles for start and

θ2A end points
Start point ‘A’
θ1A θ1B
θ1A θ2A θ3A θ1(t) θ2A θ2B
θ1B
θ3A θ3B
θ1A
Arbitrary path in
the task space
A B
Travel time T

Task Space Parametric Representation

Joint Space Planning

End point ‘B’

θ1B θ2B θ3B θ3B

θ3(t) 3. Interpolation of start and

θ2 θ3 θ3A end joint angles by
smooth functions
θ1
θ2B θ2(t) θ1(t): θ1A → θ1B
θ2(t): θ2A → θ2B
θ3(t): θ3A → θ3B
θ2A
Start point ‘A’
θ1A θ2A θ3A θ1(t)
θ1B 4. Joint space trajectories for
θ1A each joint
Arbitrary path in
the task space
A B
Travel time T

Task Space Parametric Representation

Smooth Motion Quality of Work
 Non smooth trajectories lead to low quality in
production.

Vibration

Error in path tracking

Manipulator wear

Poor quality in task

Linear Trajectory

s qg − qs
q (t ) = q + t
tg
qg − qs
q& (t ) =
tg
∞ t = 0, t g
q&&(t ) = 
0 0 < t < t g

 Infinite accelerations at endpoints

 Discontinuous velocity when two trajectory
segments are connected (at via points)
 Triangular Velocity Trajectory
tg  t 
qg − qs =  q&&min g 
2 2
4( q g − q s )
q&&min =
t g2

q s + 0.5q&&min t 2 0 ≤ t < 0.5t g

q (t ) =  s g 2
0.5(q + q ) − 0.5q&&min t 0.5t g ≤ t ≤ t g

 Acceleration discontinuity at endpoints and at

the midpoint of the trajectory

Linear Trajectory with Parabolic Blends

 Zero acceleration in middle

segments. Acceleration
discontinuity at
 Constant acceleration at end blend points
segments.
 Linear Trajectory with Parabolic Blends

 Total angular motion

S = 2( parabolic ) + Linear
1
q g − q s = 2 × q&&btb2 + q& b (t g − tb )
2
q&&btb2 − q&&bt g tb + (q g − q s ) = 0

tg1 2 4(q g − q s )
tb = ± tg −
2 2 q&&b

 For a linear part to exist

2 4( q g − q s ) b 4(q g − q s ) Minimum joint
t −
g >0 q&& >
&q&b t g2 acceleration

Linear Trajectory with Parabolic Blends

 Inclusion of Via Points into a Linear
Trajectory with Parabolic Blends

 Via points (knot points) can be introduced

between start and goal (qs, qg) positions with
constant acceleration at via point.

Multi-stage linear parabolic blend spline

Cubic Polynomial (Bring in Smoothness)

60

Joint Position Cubic 40

position

20
(1)
0
0 0.2 0.4 0.6 0.8 1
time

Joint Velocity Parabolic 100

Velocity

(2) 50

0
0 0.2 0.4 0.6 0.8 1
time
400
Joint Acceleration Linear
Acceleration

200

0
(3)
-200

-400
0 0.2 0.4 0.6 0.8 1
time
Joint Jerk -719

-719.5
(4)
Jerk

-720

-720.5

-721
0 0.2 0.4 0.6 0.8 1
time
Cubic Polynomial (zero sped at end-
points)
 Satisfies position and velocity at end-points
 Eg: zero speed at end-points
q ( 0) = qs (1) for t = 0, q s = a0
q (t g ) = qg (1) for t = t g , q g = a0 + a1t g + a2t g2 + a3t g3
q& (0) = 0 (2) for t = 0, q& s = a1
q& (t g ) = 0 (2) for t = t g , q& g = a1 + 2a2t g + 3a2t g2

3 g s 2 2 g s 3 = qs
q(t ) = q s + 2
( q − q )t − 3
( q − q )t a0
tg tg a1 = q& s (= 0)
0 ≤ t ≤ tg 3(q g − q s )
a2 =
 Acceleration is linear and uncontrollable t g2
− 2(q g − q s )
q&&(t ) = 2a2 + 6a2t 0 ≤ t ≤ t g a3 =
t g3

Cubic Polynomial (zero sped at end-

points)
Cubic Polynomial (nonzero speeds at
end- points)
 Satisfies position and velocity at end-points
 Eg: non-zero speeds at end-points
q ( 0) = qs (1) for t = 0, q s = a0
q (t g ) = q g (1) for t = t g , q g = a0 + a1t g + a2t g2 + a3t g3
q& (0) = q& s
(2) for t = 0, q& s = a1
q& (t g ) = q& g
(2) for t = t g , q& g = a1 + 2a2t g + 3a2t g2
2
q (t ) = a0 + a1t + a2t + a3t 3 a0 = qs
0 ≤ t ≤ tg a1 = q& s
3 1
a2 = 2 (q g − q s ) − (q& g − q& s )
 Acceleration is linear and tg tg
uncontrollable 2 1
a3 = − 3 (q g − q s ) + 2 (q& g − q& s )
tg tg
q&&(t ) = 2a2 + 6a2t 0 ≤ t ≤ t g

Cubic Spline Trajectory

 Stitching cubic polynomials together

 Acceleration is not continuous at via (stitching points)

5th Order Polynomial
(more oscillatory) 80

60

position
40

20

0
0 0.2 0.4 0.6 0.8 1
time
150

100

Velocity
50

0
0 0.2 0.4 0.6 0.8 1
time
400

Acceleration
200

-200

-400
0 0.2 0.4 0.6 0.8 1
time
4000

2000
Jerk

-2000
0 0.2 0.4 0.6 0.8 1
time